Power distribution automation terminal intelligent alarm system

Abstract

The invention relates to the technical field of big data analysis, in particular to a power distribution automation terminal intelligent alarm system, which comprises a multi-dimensional data fusion module for collecting equipment load rates and voltage offsets of circuit breakers and reclosers in a power grid in real time, and calculating a comprehensive vulnerability index after standardizing data from all sources. According to the method, multi-dimensional operation data such as equipment load rates and voltage offsets of the circuit breaker and the recloser are collected in real time, and data standardization and comprehensive vulnerability index calculation are combined, so that unified expression of multi-source heterogeneous characteristics of the data is effectively realized, and accurate comparability among the data participating in calculation is ensured; a power grid state reference coordinate is established based on the boundary value of the comprehensive vulnerability index, so that the state judgment of each device has a reference; and the difference between each element and an ideal operation state is described in a quantitative manner by introducing a bidirectional deviation metric value, so that the fine-grained identification precision of the equipment operation risk is improved.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, and in particular to an intelligent alarm system for power distribution automation terminals. Background Technology

[0002] Big data analytics technology refers to a comprehensive technology that involves collecting, storing, cleaning, mining, modeling, and analyzing massive, diverse, and rapidly growing data to discover hidden patterns, trends, and relationships, thereby supporting decision-making and prediction.

[0003] Current technologies rely on the inherent patterns in the data itself for macro-level statistical and trend analysis during data processing. Risk assessments of the power grid often depend on overall or average performance, neglecting the actual risk differences of individual components across various vulnerability factors. This may mask the risks of specific equipment malfunctions, leading to underreporting of risks associated with individual devices. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an intelligent alarm system for power distribution automation terminals.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A power distribution automation terminal intelligent alarm system includes:

[0006] The multi-dimensional data fusion module collects the equipment load rate and voltage deviation of each circuit breaker and recloser in the power grid in real time. After standardizing the data from all sources, it calculates the comprehensive vulnerability index and obtains the component vulnerability weighted matrix.

[0007] The vulnerability benchmark construction module, based on the component vulnerability weighting matrix, traverses all normalized values ​​of all load switches and tie switches, identifies and extracts the maximum and minimum values ​​of each normalized value, establishes vulnerability factor boundary vectors, combines all the maximum values ​​in the vulnerability factor boundary vectors into a worst-case reference point, and combines all the minimum values ​​into an optimal state reference point, thus constructing the power grid state benchmark coordinates.

[0008] The equipment status index generation module calculates the bidirectional deviation metric value of the equipment based on the component vulnerability weighting matrix and the power grid status reference coordinates.

[0009] The dynamic risk indication and alarm module filters the bidirectional deviation metric values ​​of all components in the power grid based on the bidirectional deviation metric values ​​of the equipment, forming a high-risk component sequence in the network. According to the sorting position of each component in the high-risk component sequence, it assigns different alarm color labels on the power grid wiring diagram or geographic information view, generating a dynamic risk heat map of the distribution terminal.

[0010] Preferably, the step of obtaining the component vulnerability weighting matrix is ​​as follows:

[0011] Call the real-time operation records of the circuit breakers and reclosers in the monitoring system, read the current load rate, voltage deviation, time difference between the equipment's manufacturing time and the current time, the manufacturer's fault score in the long-term operation fault statistics database, and the time difference between the last maintenance and the current time for each device, and generate a basic attribute parameter table for the components.

[0012] Based on the basic attribute parameter table of the components, calculate the comprehensive vulnerability index of each component;

[0013] The risk trend is determined based on the comprehensive vulnerability index, and the normalized values ​​of load rate, absolute value of voltage deviation, service life, manufacturer score, and maintenance interval are summarized to generate a component vulnerability weighted matrix.

[0014] Preferably, the step of obtaining the vulnerability factor boundary vector is as follows:

[0015] Based on the component vulnerability weighting matrix, the load rate normalization value, voltage offset absolute value normalization value, service life normalization value, manufacturer rating normalization value, and maintenance interval time normalization value in the component vulnerability weighting matrix are read row by row to generate a set of equipment bidirectional deviation measurement values.

[0016] Based on the set of bidirectional deviation metrics of the device, each normalized value in the set is scanned one by one, and the maximum and minimum values ​​that appear during the scanning process are recorded. The recorded maximum and minimum values ​​are then paired to form a vulnerability factor boundary vector.

[0017] Preferably, the steps for obtaining the power grid state reference coordinates are as follows:

[0018] Based on the vulnerability factor boundary vector, the maximum value of all numerical pairs in the vulnerability factor boundary vector is extracted and sequentially concatenated according to the order of the numerical pairs to form the worst-case reference point. The minimum value of all numerical pairs in the vulnerability factor boundary vector is extracted and sequentially concatenated according to the order of the numerical pairs to form the optimal reference point. The worst-case reference point and the optimal reference point are combined and paired to generate the power grid state reference coordinates.

[0019] Preferably, the step of obtaining the bidirectional deviation metric value of the device is as follows:

[0020] Based on the column vectors corresponding to the circuit breaker and recloser numbers in the component vulnerability weighted matrix, all normalized values ​​under the current timestamp are extracted according to the time series, and the index positions are aligned with the best state reference point and the worst state reference point in the power grid state reference coordinates, respectively, to generate the equipment vulnerability vector and state comparison reference set.

[0021] Based on the device vulnerability vector and state comparison reference set, calculate the bidirectional deviation metric of the device from the optimal state reference point and the worst state reference point;

[0022] Based on the bidirectional deviation metric value of the equipment, the calculation results of all circuit breakers and reclosers are iterated one by one to form a set of bidirectional deviation metric values ​​of the equipment.

[0023] Preferably, the step of obtaining the high-risk network component sequence is as follows:

[0024] Based on the set of bidirectional deviation metrics of the equipment, the bidirectional deviation metrics of all circuit breakers and reclosers in the set of bidirectional deviation metrics of the equipment are extracted, sorted in descending order by the comprehensive vulnerability index, and the bidirectional deviation metrics of each equipment and the corresponding equipment number are extracted and recorded in turn to generate a descending sorted sequence of bidirectional deviation metrics of the equipment.

[0025] Based on the descending sorted sequence of the device bidirectional deviation metric values, the total number of devices in the descending sorted sequence of device bidirectional deviation metric values ​​is counted, and 20% of the total number of devices is used as the screening threshold. Starting from the top of the sorted sequence, the device bidirectional deviation metric values ​​and corresponding device numbers are selected one by one to generate a sequence of high-risk network components.

[0026] Preferably, the steps for obtaining the dynamic risk heat map of the power distribution terminal are as follows:

[0027] Based on the network high-risk component sequence, the sorting position of each circuit breaker and recloser in the network high-risk component sequence is read one by one, and according to the correspondence between the sorting position and the preset alarm color level, the corresponding alarm color is marked on the component number position in the power grid wiring diagram or geographic information view to generate a dynamic risk heat map of the power distribution terminal.

[0028] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0029] In this invention, multi-dimensional operational data such as equipment load rate and voltage deviation of circuit breakers and reclosers are collected in real time. Combined with data standardization and comprehensive vulnerability index calculation, the unified expression of multi-source heterogeneous characteristics of data is effectively realized, ensuring the accurate comparability between the data involved in the calculation. Furthermore, a grid state benchmark coordinate is established based on the boundary value of the comprehensive vulnerability index, providing a benchmark reference for judging the state of each device. By introducing bidirectional deviation metrics, the gap between each component and the ideal operating state is described quantitatively, improving the fine-grained identification accuracy of equipment operation risks. Finally, by combining dynamic risk ranking and spatial mapping operations, the distribution of network risks is displayed intuitively in a visual manner, realizing real-time dynamic monitoring and location of risks. This improves the immediacy and accuracy of distribution network risk warning, shortens risk response delay, and enhances the overall grid operation safety. Attached Figure Description

[0030] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0032] Please see Figure 1 The present invention provides a technical solution: a power distribution automation terminal intelligent alarm system comprising:

[0033] The multi-dimensional data fusion module collects the equipment load rate and voltage deviation of each circuit breaker and recloser in the power grid in real time. After standardizing the data from all sources, it calculates the comprehensive vulnerability index and obtains the component vulnerability weighted matrix.

[0034] The vulnerability benchmark construction module, based on the component vulnerability weighting matrix, traverses all normalized values ​​of all load switches and tie switches, identifies and extracts the maximum and minimum values ​​of each normalized value, establishes vulnerability factor boundary vectors, combines all the maximum values ​​in the vulnerability factor boundary vectors into the worst-case reference point, and combines all the minimum values ​​into the best-case reference point, thus constructing the power grid state benchmark coordinates.

[0035] The equipment status index generation module calculates the bidirectional deviation metric of the equipment based on the component vulnerability weighting matrix and the power grid status reference coordinates.

[0036] The dynamic risk indication and alarm module filters the bidirectional deviation values ​​of all components in the power grid based on the bidirectional deviation values ​​of the equipment, forming a sequence of high-risk components in the network. According to the sorting position of each component in the sequence of high-risk components in the network, it assigns different alarm color labels on the power grid wiring diagram or geographic information view, and generates a dynamic risk heat map of the distribution terminal.

[0037] The steps to obtain the component vulnerability weighting matrix are as follows:

[0038] Call the real-time operation records of the circuit breakers and reclosers in the monitoring system, read the current load rate, voltage deviation, time difference between the equipment's manufacturing time and the current time, the manufacturer's fault score in the long-term operation fault statistics database, and the time difference between the last maintenance and the current time for each device, and generate a basic attribute parameter table for the components.

[0039] Based on the component basic attribute parameter table, calculate the comprehensive vulnerability index for each component. The calculation formula is as follows:

[0040]

[0041] Among them, S i Let i be the overall vulnerability index of the i-th component.

[0042] Let L be the normalized load rate value of the i-th element. i For equipment load rate, L min L max These are the minimum and maximum load rates in the entire sample, respectively.

[0043] Q is the normalized absolute value of the voltage offset of the i-th element. i For voltage offset, |Q| min |Q| max These are the minimum and maximum absolute values ​​of the voltage offset for all devices.

[0044] Let A be the normalized value of the service life of the i-th component. i The service life of the equipment from the date of manufacture to the present, A min A max These represent the shortest and longest service life in the sample, respectively.

[0045] M is the inversely normalized value of the score for the i-th component manufacturer. i M is the manufacturer's historical average failure score. min M max The extreme values ​​of the scoring interval.

[0046] R is the normalized value of the maintenance interval time for the i-th component. i R represents the number of days since the last maintenance. min R max These are the shortest and longest intervals in the sample.

[0047] w k Let w be the linear weighting coefficient of the normalized value of the k-th term, k∈{L,Q,A,M,R}, satisfying ∑w k =1-w int ,

[0048] w int This is the adjustment coefficient for the nonlinear coupling term, used to adjust the combined effect of high load rate and high aging. (Nonlinear term) For modeling when the equipment load rate P′ i,L and service life P′ i,A Simultaneously, the degradation effect index increases during the rise, with a value range of [-1, e-1);

[0049] The risk trend is determined based on the comprehensive vulnerability index, and the normalized values ​​of load rate, absolute value of voltage deviation, service life, manufacturer score, and maintenance interval are summarized to generate a component vulnerability weighted matrix.

[0050] Specifically, the system accesses the real-time operation records of circuit breakers and reclosers from the monitoring system. This is achieved by sending data requests to the SCADA or DMS system of the distribution network to obtain the operational data stream of each circuit breaker and recloser at the current timestamp. The system then analyzes and extracts various parameters from the data stream, starting with the load rate, which is determined by reading the real-time current value I. realtime Divide by the rated operating current I specified on the equipment nameplate. rated The equipment load rate L was calculated. i =(I realtime / I rated For example, if a circuit breaker with a rated current of 630A has a real-time monitored current of 410A, then its load factor is (410 / 630)×100%≈65.1%. Next is the voltage offset, which is determined by reading the real-time voltage value U. realtime and the line's rated voltage U rated Compare and calculate the voltage offset Q i =((U) realtime -U rated ) / U ratedFor example, on a 10kV line, if a recloser monitors a voltage of 10.3kV, then the voltage deviation is ((10.3-10) / 10)×100%=3%. Next, the time difference between the equipment's manufacturing date and the current time is used as the equipment's service life. This is obtained by reading the manufacturing date from the equipment asset management database and subtracting it from the current system date. The result is recorded in years. Then, there is the manufacturer's fault score, which is obtained through in-depth analysis of the historical operational fault statistics database. The specific calculation method is as follows: Where n s w represents the number of times the manufacturer's equipment experienced a severity level of s failure over the past five years. s Y represents the weight corresponding to the severity level of the fault. total The total operating years of all equipment from this manufacturer are used as the basis for assigning three levels of severity weights to the faults: Level 1 is a transient fault that the equipment can recover on its own, with a weight w1 = 1; Level 2 is a permanent fault that requires manual intervention, with a weight w2 = 5; and Level 3 is a severe fault that causes power outages for downstream users, with a weight w3 = 10. For example, if a manufacturer's equipment has a total operating life of 2000 years and has experienced 50 Level 1 faults, 10 Level 2 faults, and 2 Level 3 faults, then its fault score is (50·1 + 10·5 + 2·10) / 2000 = 0.06. Finally, the time difference between the last maintenance and the current time is calculated by accessing the equipment maintenance record database to extract the date of the last preventive test or maintenance work and calculate the time interval in days between the last and the current date. The above five key data items—equipment load rate, voltage deviation, equipment service life, manufacturer fault score, and maintenance interval—along with the equipment's unique identifier (ID), are systematically organized and stored in a structured data table to generate a component basic attribute parameter table.

[0051] formula: The advantage of this formula lies in its combination of a linear weighted model and a nonlinear coupling effect model, which provides a more accurate characterization of the vulnerability of distribution terminals. The linear part... It can comprehensively and evenly assess fundamental risks from multiple dimensions, including equipment operating conditions, physical aging, manufacturer quality, and maintenance status, rather than nonlinear terms. This method quantifies the synergistic degradation effect caused by the simultaneous action of two stress factors: high load and aging. It simulates the reality of a sharp increase in risk when "old equipment is operated under heavy load," and adjusts the coefficient w. int The contribution of this coupling effect to the overall vulnerability index can be flexibly adjusted, enabling the model to adapt to risk preferences under different types of power grids and operation and maintenance strategies.

[0052] P′ i,kThis is the normalized value of the k-th vulnerability factor for the i-th component, encompassing five indicators (k∈{L,Q,A,M,R}): load factor, voltage offset, service life, manufacturer rating, and maintenance interval. These normalized values ​​are calculated using the max-min normalization method to ensure all factors are within the [0,1] interval for easier subsequent weighted calculations. The calculations are based on the component basic attribute parameter table generated in the previous step and statistical analysis of sample data from all similar devices across the entire network. For example, the load factor normalized value P′... i,L The calculation formula is: Where L i Let L be the current load rate of device i, and L be the load rate of device i. min With L max This is achieved by iterating through the load rates of all devices in the component's basic attribute parameter table and finding the minimum and maximum values. For example, after statistically analyzing 1000 circuit breakers across the entire network, the load rate range is found to be between 10% and 98%, i.e., L... min =0.1, L max =0.98, if the current load factor L of a certain circuit breaker i is 0.98 i If it is 85%, then its normalized value is... The remaining four normalized values ​​P′ i,Q 、P′ i,A 、P′ i,M 、P′ i,R A similar calculation method is used, but attention should be paid to the manufacturer rating P′. i,M Inverse normalization is used to ensure that a higher score represents a greater risk.

[0053] w k The linear weighting coefficient for the k-th normalized value is set based on expert evaluations of the contribution of each vulnerability factor to equipment failure. A 5x5 judgment matrix is ​​constructed, and five senior power grid operation and maintenance experts compare the five vulnerability factors (load rate L, voltage deviation Q, service life A, manufacturer M, and maintenance interval R) pairwise. Based on their relative importance, they are scored using a 1-9 scale. For example, if experts generally consider load rate (L) to be "slightly more important" than voltage deviation (Q), then LQ is scored as 3, while QL is scored as 1 / 3. After collecting the judgment matrices from all experts, the geometric mean of each matrix is ​​calculated to obtain the final consensus judgment matrix. Then, the largest eigenvalue and its corresponding eigenvector of this matrix are calculated, and the eigenvector is normalized to obtain the initial weights of each factor. For example, the initial weight vector is calculated as [0.45, 0.10, 0.25, 0.10, 0.10], corresponding to w L ,w Q ,w A ,w M ,w RFinally, based on the weight w of the nonlinear term... int Make adjustments so that ∑w k =1-w int If w int If we set it to 0.2, then the final linear weight coefficient is the initial weight multiplied by (1-0.2) = 0.8, i.e., w. L =0.36,w Q =0.08,w A =0.20,w M =0.08,w R =0.08.

[0054] w int The adjustment coefficient for the nonlinear coupling term is used to control the impact of the combined effects of high load rate and high aging on the overall vulnerability index. Its value is determined based on regression analysis of historical fault data. First, archival data of all circuit breakers and reclosers that have failed in the power grid over the past ten years are collected, and the normalized load rate value P′ at the moment before the fault is extracted. i,L and the normalized value of service life P′ i,A The variables are assigned values ​​and their failure status is marked (failure = 1, no failure = 0). Then, a logistic regression model is constructed, where the dependent variable is whether a failure has occurred and the independent variable is the linear vulnerability component ∑w. k ·P′ i,k and nonlinear terms By fitting the model, the optimal regression coefficient for the nonlinear term can be obtained, and this coefficient is the adjustment coefficient w. int For example, based on logistic regression analysis of tens of thousands of historical data points, the coefficient of the nonlinear term output by the model is 0.213. This coefficient value is rounded down and set to w. int =0.2, which means that in the comprehensive vulnerability assessment, the coupling effect of load and aging accounts for 20% of the weight, while the linear combination of the other five basic factors accounts for 80% of the weight.

[0055] Calculation process:

[0056] Taking a circuit breaker numbered CB-101 in the network as an example, calculate its comprehensive vulnerability index S. 101 First, obtain the original data and extreme values ​​of the entire network samples from the component's basic attribute parameter table:

[0057] Equipment load rate L 101 =82%, total network sample L min =10%, L max =98%.

[0058] Voltage offset Q 101 = -4.5%, absolute value of voltage deviation of the entire network sample |Q| min=0.1%,|Q| max =6.0%.

[0059] Service life A 101 =18 years, full network sample A min =1 year, A max =25 years.

[0060] Manufacturer Fault Score M 101 =0.8, rating interval M min =0.05,M max =1.2.

[0061] Maintenance interval R 101 =700 days, total network sample R min =30 days, R max = 1100 days.

[0062] Based on the above data, calculate the normalized values ​​for each item:

[0063]

[0064] The weighting coefficients determined using the aforementioned steps are as follows:

[0065] w L =0.36,w Q =0.08,w A =0.20,w M =0.08,w R =0.08, and w int =0.2.

[0066] Substitute the normalized values ​​and weighting coefficients into the formula for calculation:

[0067] Linear part calculation:

[0068] ∑w k ·P′ 101,k = 0.36·0.818+0.08·0.746+0.20·0.708+0.08·0.348+0.08·0.626;

[0069] =0.29448+0.05968+0.1416+0.02784+0.05008=0.57368;

[0070] Nonlinear calculation:

[0071] P′ 101,L +P′ 101,A =0.818 + 0.708 = 1.526;

[0072]

[0073] Comprehensive vulnerability index calculation:

[0074] S 101 =0.57368 + 0.1384 = 0.71208;

[0075] The result indicates that the comprehensive vulnerability index of circuit breaker CB-101 is 0.71208. This value is a comprehensive risk measure that quantifies the multi-dimensional health status of the equipment into a single, comparable indicator. According to the preset risk classification standard, for example, [0,0.3) is low risk, [0.3,0.7) is medium risk, and [0.7,1.0] is high risk. This classification standard is based on the statistical analysis of the historical vulnerability index distribution of all network equipment, and the 90th percentile is set as the high risk threshold. Therefore, the result of 0.71208 indicates that the equipment has entered the high-risk range, and its operating status deserves high attention.

[0076] Based on the comprehensive vulnerability index of each component calculated in the previous steps, the system first initiates the risk trend assessment process. This process calculates the comprehensive vulnerability index S of each component. i Compared with a preset risk threshold, the risk threshold is set based on statistical analysis of the long-term distribution of historical vulnerability indices of all network devices. Specifically, all historical index values ​​are sorted, and the 70th and 95th percentiles are used as the dividing lines for risk levels. For example, if the calculated 70th percentile value is 0.45 and the 95th percentile value is 0.72, then the risk level is defined as: S i <0.45 indicates "low risk", 0.45≤S i <0.72 indicates "medium risk", S i A value ≥0.72 is considered "high risk." This judgment is temporarily attached to the data record of each component. Subsequently, the system performs a data aggregation operation, normalizing the five key vulnerability factors of each circuit breaker and recloser, i.e., the load rate normalized value P′, to construct the matrix required for subsequent analysis. i,L The normalized absolute value of voltage offset P′ i,Q Normalized value of service life P′ i,A The inverse normalized value P′ of the manufacturer rating i,M and the normalized value of maintenance interval P′ i,R Extracted from the computation cache, these normalized values ​​represent the state of the device relative to the entire group in various dimensions, forming the basis for multi-dimensional comparison and analysis. The system integrates these data with the unique identifier of the component (such as the device ID) and the risk level label that was just determined. Finally, it organizes this information of all components into a two-dimensional table-like data structure to generate a component vulnerability weighted matrix.

[0077] The steps to obtain the vulnerability factor boundary vector are as follows:

[0078] Based on the component vulnerability weighting matrix, the load rate normalization value, voltage offset absolute value normalization value, service life normalization value, manufacturer rating normalization value, and maintenance interval normalization value are read row by row in the component vulnerability weighting matrix to generate a set of equipment bidirectional deviation measurement values.

[0079] Based on the set of bidirectional deviation metrics for the device, each normalized value in the set is scanned one by one, and the maximum and minimum values ​​that appear during the scanning process are recorded. The recorded maximum and minimum values ​​are then paired to form a vulnerability factor boundary vector.

[0080] Specifically, based on the component vulnerability weighted matrix, the system initiates a data extraction program. This program iteratively processes the matrix row by row. Since each row of the matrix represents a unique circuit breaker or recloser device and contains all its vulnerability characteristics, the program locates and reads five predefined key data columns within that row in each iteration. These five data columns are the load rate normalized value, the absolute value of the voltage offset normalized value, the service life normalized value, the manufacturer's rating reverse normalized value, and the maintenance interval normalized value. For the first row of the matrix, representing the circuit breaker with device ID "CB-001", the program extracts its corresponding five normalized values, for example, [0.75, 0.21, 0.88, 0.45, 0.62]. These five values ​​constitute a... The five-dimensional vector fully describes the vulnerability profile of the device at the current moment. The program appends the device ID "CB-001" as an index to this vector, and then stores this indexed vector in a temporary dynamic array. Next, the program automatically moves to the second row of the matrix and performs the same operation on the device "RC-002", extracting its five-dimensional vector [0.43, 0.55, 0.34, 0.71, 0.29] and storing it in the dynamic array. This process will be repeated continuously until all rows of the component vulnerability weighting matrix have been traversed, and the vulnerability state of each circuit breaker and recloser in the network is transformed into a standardized five-dimensional vector. Finally, this dynamic array containing the five-dimensional vulnerability vectors of all devices in the network is solidified, generating a set of device bidirectional deviation metrics.

[0081] Based on the set of bidirectional deviation metrics generated in the previous process, the system initializes five pairs of variables to store extreme values, corresponding to five vulnerability factors, specifically (L... min_val ,L max_val ), (Q min_val Q max_val ), (A min_val A max_val), (M min_val M max_val ) and (R min_val ,R max_val ), where all minimum value variables (such as L) min_val ) is assigned an initial value of 1.0, and all variables with maximum values ​​(such as L) max_val The initial value is assigned to 0.0. Subsequently, the system initiates a comprehensive scan. This scan first identifies the first vulnerability factor, namely load rate. Then, it iterates through the five-dimensional vector of each device in the set of device bidirectional deviation metrics, reading only the first element, namely the load rate normalization value. For example, when the first device's load rate normalization value of 0.75 is read, the program will compare it with the current L. min_val (1.0) and L max_val Comparing (0.0), since 0.75 is less than 1.0, L min_val Updated to 0.75, and since 0.75 is greater than 0.0, L max_val It was also updated to 0.75. When the normalized load rate value of the next device was read as 0.43, the program compared again, L. min_val It was updated to 0.43, while L max_val Keeping the value constant at 0.75, this process continues until the load rate normalization values ​​of all devices in the set have been traversed. At this point, the obtained L... min_val and L max_val This refers to the actual minimum and maximum values ​​of the normalized load rate among all network devices. Next, the program scans and updates the extreme values ​​of the normalized values ​​for voltage offset, service life, manufacturer rating, and maintenance interval in the same manner. After scanning all five factors, the system records five sets of maximum and minimum values, i.e., (L... min_val ,L max_val ), (Q min_val Q max_val These, along with others, form numerical pairs to create the vulnerability factor boundary vector.

[0082] The steps for obtaining the power grid state reference coordinates are as follows:

[0083] Based on the vulnerability factor boundary vector, the maximum value of all numerical pairs in the vulnerability factor boundary vector is extracted and then concatenated in the order of the numerical pairs to form the worst-case reference point. The minimum value of all numerical pairs in the vulnerability factor boundary vector is extracted and then concatenated in the order of the numerical pairs to form the optimal reference point. The worst-case reference point and the optimal reference point are then combined and paired to generate the power grid state reference coordinates.

[0084] Specifically, based on the vulnerability factor boundary vector, the system begins to construct two core state reference points. First, it forms the worst-case state reference point. The program accesses the vulnerability factor boundary vector, which consists of five numerical pairs, for example, [(0.12,0.95),(0.05,0.89),(0.02,0.98),(0.10,0.91),(0.08,0.96)], corresponding to the (minimum, maximum) pairs of load factor, voltage offset, service life, manufacturer rating, and maintenance interval, respectively. The program extracts the maximum value from each numerical pair sequentially according to a preset factor order: load factor, voltage offset, service life, manufacturer rating, and maintenance interval. Specifically, it extracts 0.95 from the first numerical pair (0.12,0.95), 0.89 from the second numerical pair (0.05,0.89), and so on. Similarly, the process continues until 0.96 is extracted from the last numerical pair (0.08, 0.96). Then, the program sequentially concatenates these five extracted maximum values ​​[0.95, 0.89, 0.98, 0.91, 0.96] to form a five-dimensional vector, which is the worst-case reference point. Next, the program uses the same logic to form the optimal reference point. It again traverses the vulnerability factor boundary vector, but this time extracts the minimum value from each numerical pair, that is, extracts 0.12, 0.05, 0.02, 0.10, 0.08 in sequence, and concatenates them sequentially to form another five-dimensional vector [0.12, 0.05, 0.02, 0.10, 0.08]. This vector is the optimal reference point. Finally, these two reference point vectors, namely the worst-case reference point and the optimal reference point, are combined and paired as a whole to generate the power grid state reference coordinates.

[0085] The steps for obtaining the bidirectional deviation measurement value of the equipment are as follows:

[0086] Based on the column vectors corresponding to the circuit breaker and recloser numbers in the component vulnerability weighted matrix, all normalized values ​​under the current timestamp are extracted according to the time series, and the index positions are aligned with the best state reference point and the worst state reference point in the power grid state reference coordinates, respectively, to generate the equipment vulnerability vector and state comparison reference set.

[0087] Based on the equipment vulnerability vector and the state comparison reference set, the bidirectional deviation metric of the equipment from the optimal state reference point and the worst state reference point is calculated using the following formula:

[0088]

[0089] Among them, D c P′ represents the bidirectional deviation metric of the c-th circuit breaker or recloser element. c,g U represents the standardized value of the c-th device on the g-th vulnerability factor. gW represents the value of the optimal state reference point on the g-th factor. g β represents the value of the worst-case reference point on the g-th factor. g Let represent the distance sensitivity factor of the g-th factor, p be the norm coefficient that controls the calculation method of the distance family, and z be the total number of vulnerability factors.

[0090] Based on the bidirectional deviation metric of the equipment, the calculation results of all circuit breakers and reclosers are iterated one by one to form a set of bidirectional deviation metric values ​​of the equipment.

[0091] Specifically, based on the component vulnerability weighting matrix, the system first starts an independent instance processing thread for each circuit breaker and recloser in the network to extract their respective state vectors in a parallel manner. For a specific device, such as circuit breaker numbered CB-101, the system locks the row corresponding to that device in the component vulnerability weighting matrix. This row contains all the normalized vulnerability factors of the device at the current timestamp, specifically the normalized values ​​of load rate, absolute voltage offset, service life, manufacturer rating, and maintenance interval. The system extracts these five values ​​in a preset order [load rate, voltage offset, service life, manufacturer, maintenance] to form a five-dimensional device vulnerability vector. For example, the extracted vector is [0.818, 0.746, 0.708, 0.348, 0.626]. At the same time, the system retrieves the already... The power grid state baseline coordinates constructed in the previous steps contain the optimal state reference point vector, such as [0.12, 0.05, 0.02, 0.10, 0.08], and the worst-case state reference point vector, such as [0.95, 0.89, 0.98, 0.91, 0.96]. Next, the system performs an alignment operation, matching the vulnerability vector of device CB-101 with these two baseline reference point vectors to ensure that the five dimensions (i.e., the five vulnerability factors) of the three vectors correspond one-to-one at the index positions. Finally, the system packages these three vectors (the device's own vulnerability vector, the optimal state reference point, and the worst-case state reference point) into a data structure and indexes it with the device number CB-101 as the key. This data package containing the three vectors is the comparison reference unit for a single device. This process is executed synchronously for all devices, generating a set of device vulnerability vectors and state comparison references.

[0092] formula: The advantage of this formula lies in its ability to achieve a refined and relative assessment of equipment risk levels by calculating the weighted distance difference between the current state of the equipment and the two extreme benchmarks of the power grid's "optimal" and "worst" states. It not only measures the degree to which the equipment deviates from its ideal state (first term), but more importantly, it also considers the degree to which it approaches the most dangerous state (second term). Positive values ​​indicate that the equipment is closer to the worst state, while negative values ​​indicate that it is closer to the optimal state. The absolute value reflects the severity of the deviation. Compared to the traditional one-way threshold judgment method, this can more effectively identify equipment that is in the "middle ground" but has shown a clear trend of deterioration. Furthermore, by introducing a distance sensitivity factor β... g It allows operations and maintenance managers to assign different weights to different vulnerability factors based on their impact on equipment security.

[0093] P′ c,g U g and W g These are, respectively, the standardized value of the c-th device on the g-th vulnerability factor, the value of the best-case reference point on the g-th factor, and the value of the worst-case reference point on the g-th factor. These three parameters are directly derived from the device vulnerability vector and state comparison reference set generated in the previous step and do not need to be recalculated. Where, P′ c,g It is the core data describing the current state of the device, and is a normalized value in the interval [0,1]. g and W g Together, they constitute the "benchmark" for the evaluation, representing the best and worst levels that all network devices can achieve on this factor. For example, for device CB-101, its device vulnerability vector P′ 101 The values ​​are [0.818, 0.746, 0.708, 0.348, 0.626], while the optimal state reference point U obtained from the power grid state reference coordinates is [0.12, 0.05, 0.02, 0.10, 0.08], and the worst state reference point W is [0.95, 0.89, 0.98, 0.91, 0.96]. In subsequent calculations, these values ​​will be used one by one according to their index positions.

[0094] β gLet g represent the distance sensitivity factor of the g-th factor. It reflects the degree of importance attached to the deviations of different vulnerability factors when measuring equipment status deviation. A review panel of 10 experts with over 15 years of experience in distribution network operation and maintenance was organized. The experts were required to compare the importance of five vulnerability factors (load rate, voltage deviation, service life, manufacturer, and maintenance interval) pairwise based on the criterion of "the direct contribution of equipment status deviation to the occurrence of faults," constructing a judgment matrix. For example, if the expert panel unanimously agreed that the deviation of load rate was "strongly important" in contributing to risk than the deviation of voltage deviation, then an evaluation value of 7 would be assigned. After collecting the judgment matrices of all experts, their geometric mean was calculated to form a comprehensive judgment matrix. Then, the maximum eigenvalue and corresponding eigenvector of this matrix were calculated, and the eigenvector was normalized to obtain the weight of each factor, i.e., β. g The value of β, for example, after calculation, the sensitivity factor of each factor is obtained as: β L =0.45, β Q =0.05, β A =0.30, β M =0.05, β R =0.15.

[0095] p is the norm coefficient, used to control the method of distance calculation. Its selection determines the definition of distance in multidimensional space. When p=1, the Manhattan distance is calculated, which is the sum of the absolute values ​​of the differences between the coordinate axes. When p=2, the Euclidean distance is calculated, which is the straight-line distance between two points in space. In this system, p=2 is set and the Euclidean distance is selected. The reason is that the Euclidean distance can comprehensively consider the deviations on all vulnerability factors, smoothly reflect the overall changes in equipment status, and avoid the distortion of distance measurement caused by drastic fluctuations of a single factor.

[0096] z represents the total number of vulnerability factors. A total of 5 vulnerability factors were selected, therefore z = 5.

[0097] Calculation process:

[0098] Continuing with equipment CB-101 as an example, calculate its bidirectional deviation metric D. 101 The parameters used are as follows:

[0099] Device vulnerability vector P′ 101 =[0.818,0.746,0.708,0.348,0.626].

[0100] The optimal state reference point U = [0.12, 0.05, 0.02, 0.10, 0.08].

[0101] The worst-case reference point is W = [0.95, 0.89, 0.98, 0.91, 0.96].

[0102] Distance sensitivity factor β = [0.45, 0.05, 0.30, 0.05, 0.15].

[0103] The norm coefficient p = 2, and the total number of vulnerability factors z = 5.

[0104] Step 1: Calculate the weighted distance to the optimal state reference point U:

[0105]

[0106] Step 2: Calculate the weighted distance to the worst-case reference point W:

[0107]

[0108] Step 3: Calculate the device's bidirectional deviation metric D 101 :

[0109] D 101 =0.6582 - 0.2520 = 0.4062;

[0110] The results show that the bidirectional deviation metric of device CB-101 is 0.4062, which is a positive value. This clearly indicates that the device is closer to the "worst-case reference point" representing danger in the multidimensional vulnerability space, rather than the "optimal-case reference point" representing health. The value of 0.4062 quantifies the magnitude of this bias. The larger the value, the worse the condition and the higher the risk. This value will serve as the core quantitative indicator for risk ranking and alarm classification of devices.

[0111] Based on the bidirectional deviation metric calculated for each device in the previous steps, the system then executes a data aggregation program. The goal of this program is to integrate the scattered calculation results into a unified, ordered dataset. The program first initializes an empty data list, designed to store pairs of values ​​consisting of the device number and its corresponding bidirectional deviation metric. Then, the program initiates a traversal loop. This loop targets the previously generated memory area containing the calculation results for all devices. For each iteration of the loop, the program retrieves the calculation result for one device. For example, it first processes the circuit breaker CB-001, calculating its bidirectional deviation... When the metric value is -0.253, the program adds the data pair ("CB-001", -0.253) to the data list. Then, it iterates to the next device, recloser RC-001, whose metric value is 0.189. The program also adds ("RC-001", 0.189) to the list. This process continues uninterruptedly and one by one until the bidirectional deviation metric values ​​of all circuit breakers and reclosers in the network have been accurately read and added to the list. Finally, when the traversal loop ends, this data list contains the complete calculation results of all target devices in the entire network. The program then solidifies this list to form a set of bidirectional deviation metric values ​​for the devices.

[0112] The steps for obtaining high-risk network component sequences are as follows:

[0113] Based on the set of bidirectional deviation metrics of equipment, the bidirectional deviation metrics of all circuit breakers and reclosers in the set of bidirectional deviation metrics of equipment are extracted, sorted in descending order by the comprehensive vulnerability index, and the bidirectional deviation metrics of each equipment and the corresponding equipment number are extracted and recorded in turn to generate a descending sorted sequence of bidirectional deviation metrics of equipment.

[0114] Based on the descending sort sequence of device bidirectional deviation metric values, the total number of devices in the descending sort sequence is counted. Then, using 20% ​​of the total number of devices as the screening threshold, the bidirectional deviation metric values ​​and corresponding device numbers of devices are selected one by one from the top of the sort sequence to generate a sequence of high-risk network components.

[0115] Specifically, based on the set of device bidirectional deviation metrics, the system first extracts the bidirectional deviation metrics of all devices in the set and correlates them with the previously calculated comprehensive vulnerability index of each device. The correlation is based on the unique ID number of each device, forming a temporary data table containing three columns: device ID, bidirectional deviation metric, and comprehensive vulnerability index. Next, the system performs a dual sorting logic on this table. The primary sorting key is set to the device bidirectional deviation metric, arranged in descending order. This means that devices with larger metric values ​​(i.e., those closer to the worst-case scenario) will be ranked higher. The secondary sorting key is set to the comprehensive vulnerability index, also arranged in descending order. The purpose of this secondary sorting key is to prioritize devices with higher overall vulnerability indices when multiple devices have the same bidirectional deviation metric value. For example, if devices A and B both have a bidirectional deviation metric value of 0.51, but A has an overall vulnerability index of 0.78 and B has 0.75, then after sorting, A will be ranked before B. The sorting algorithm uses a stable quicksort algorithm. After sorting, the system will extract the bidirectional deviation metric value and its corresponding device number for each device from the sorted list in a top-to-bottom order, and store this information in a new, ordered sequence structure, generating a descending sorted sequence of device bidirectional deviation metrics.

[0116] Based on the descending sorting sequence of equipment bidirectional deviation metrics, the system first performs a counting operation. By reading the sequence length, it determines the total number of monitored circuit breakers and reclosers in the current power grid. For example, if the sequence contains 1500 equipment entries, the total number of equipment is 1500. Subsequently, the system calculates a screening threshold based on this total. This threshold is fixed at 20% of the total number of equipment. In this example, the screening threshold is calculated as 1500 × 20% = 300. This percentage is determined based on industry best practices for managing critical equipment and after Pareto analysis of historical data. The analysis shows that approximately 20% of the equipment contributes 80% of the operational costs. Therefore, the focus is on the top 20% of devices due to the risk. Next, the system starts a selection process that selects items one by one from the top of the descending sequence of devices with the highest risk, starting with the devices with the highest bidirectional deviation metric. Each time an item is selected, its bidirectional deviation metric and the corresponding device number are copied to a new list. This process continues until the number of selected devices reaches the previously calculated screening threshold of 300. When the 300th device is selected and added to the new list, the selection process stops immediately. This newly generated list, which contains information on the top 20% of the network's riskiest devices, is the sequence of high-risk network components.

[0117] The steps for obtaining the dynamic risk heat map of the power distribution terminal are as follows:

[0118] Based on the network high-risk component sequence, the sorting position of each circuit breaker and recloser in the network high-risk component sequence is read one by one, and according to the correspondence between the sorting position and the preset alarm color level, the corresponding alarm color is marked on the component number position in the power grid wiring diagram or geographic information view to generate a dynamic risk heat map of the power distribution terminal.

[0119] Specifically, based on the sequence of high-risk network components, the system launches a visualization rendering program. This program reads each device entry in the sequence, obtaining the ranking position information of each high-risk device. Simultaneously, the system loads a preset alarm color level correspondence table. This table defines the mapping rules between the risk ranking position and the display color. This mapping relationship is designed based on best practices for risk management visualization. For example, devices ranked in the top 5% of the sequence (ranked 1 to 15 in this example) are defined as "Level 1 Alarm," corresponding to dark red, indicating the highest level of emergency risk. Devices ranked between 5% and 10% (ranked 16 to 30) are designated as "Level 2 Alarm," corresponding to orange, indicating high concern. Devices ranked between 10% and 20% (ranked 31 to 60) are designated as "Level 2 Alarm." The device is designated as "Level 3 Alarm," corresponding to yellow, indicating a general warning. The remaining devices in the high-risk network component sequence (ranked 61 to 300) are uniformly marked as "Level 4 Attention," corresponding to blue. Subsequently, the rendering program binds these color information with the device IDs and calls the power grid's graphical display interface. On the main wiring diagram or GIS geographic information view, it locates the icon or symbol of each high-risk device on the map according to its ID and modifies its color attribute to the color corresponding to the alarm level. For example, the circuit breaker symbol corresponding to the first-ranked device CB-058 in the sequence will be dynamically filled with dark red on the wiring diagram. By performing this operation on all devices in the high-risk network component sequence, a dynamic risk heat map of the distribution terminals that intuitively reflects the risk distribution status of the entire network is finally generated on the screen.

Claims

1. A power distribution automation terminal intelligent alarm system, characterized in that, The system includes: The multi-dimensional data fusion module collects the equipment load rate and voltage deviation of each circuit breaker and recloser in the power grid in real time. After standardizing the data from all sources, it calculates the comprehensive vulnerability index and obtains the component vulnerability weighted matrix. The vulnerability benchmark construction module, based on the component vulnerability weighting matrix, traverses all normalized values ​​of all load switches and tie switches, identifies and extracts the maximum and minimum values ​​of each normalized value, establishes vulnerability factor boundary vectors, combines all the maximum values ​​in the vulnerability factor boundary vectors into a worst-case reference point, and combines all the minimum values ​​into an optimal state reference point, thus constructing the power grid state benchmark coordinates. The equipment status index generation module calculates the bidirectional deviation metric value of the equipment based on the component vulnerability weighting matrix and the power grid status reference coordinates. The dynamic risk indication and alarm module filters the bidirectional deviation metric values ​​of all components in the power grid based on the bidirectional deviation metric values ​​of the equipment, forming a high-risk component sequence in the network. According to the sorting position of each component in the high-risk component sequence, it assigns different alarm color labels on the power grid wiring diagram or geographic information view, generating a dynamic risk heat map of the distribution terminal.

2. The intelligent alarm system for power distribution automation terminals according to claim 1, characterized in that, The steps for obtaining the component vulnerability weighting matrix are as follows: Call the real-time operation records of the circuit breakers and reclosers in the monitoring system, read the current load rate, voltage deviation, time difference between the equipment's manufacturing time and the current time, the manufacturer's fault score in the long-term operation fault statistics database, and the time difference between the last maintenance and the current time for each device, and generate a basic attribute parameter table for the components. Based on the basic attribute parameter table of the components, calculate the comprehensive vulnerability index of each component; The risk trend is determined based on the comprehensive vulnerability index, and the normalized values ​​of load rate, absolute value of voltage deviation, service life, manufacturer score, and maintenance interval are summarized to generate a component vulnerability weighted matrix.

3. The intelligent alarm system for power distribution automation terminals according to claim 1, characterized in that, The steps for obtaining the vulnerability factor boundary vector are as follows: Based on the component vulnerability weighting matrix, the load rate normalization value, voltage offset absolute value normalization value, service life normalization value, manufacturer rating normalization value, and maintenance interval time normalization value in the component vulnerability weighting matrix are read row by row to generate a set of equipment bidirectional deviation measurement values. Based on the set of bidirectional deviation metrics of the device, each normalized value in the set is scanned one by one, and the maximum and minimum values ​​that appear during the scanning process are recorded. The recorded maximum and minimum values ​​are then paired to form a vulnerability factor boundary vector.

4. The intelligent alarm system for power distribution automation terminals according to claim 1, characterized in that, The steps for obtaining the power grid state reference coordinates are as follows: Based on the vulnerability factor boundary vector, the maximum value of all numerical pairs in the vulnerability factor boundary vector is extracted and sequentially concatenated according to the order of the numerical pairs to form the worst-case reference point. The minimum value of all numerical pairs in the vulnerability factor boundary vector is extracted and sequentially concatenated according to the order of the numerical pairs to form the optimal reference point. The worst-case reference point and the optimal reference point are combined and paired to generate the power grid state reference coordinates.

5. The intelligent alarm system for power distribution automation terminals according to claim 1, characterized in that, The steps for obtaining the bidirectional deviation metric value of the device are as follows: Based on the column vectors corresponding to the circuit breaker and recloser numbers in the component vulnerability weighted matrix, all normalized values ​​under the current timestamp are extracted according to the time series, and the index positions are aligned with the best state reference point and the worst state reference point in the power grid state reference coordinates, respectively, to generate the equipment vulnerability vector and state comparison reference set. Based on the device vulnerability vector and state comparison reference set, calculate the bidirectional deviation metric of the device from the optimal state reference point and the worst state reference point; Based on the bidirectional deviation metric value of the equipment, the calculation results of all circuit breakers and reclosers are iterated one by one to form a set of bidirectional deviation metric values ​​of the equipment.

6. The intelligent alarm system for power distribution automation terminals according to claim 1, characterized in that, The steps for obtaining the high-risk component sequence of the network are as follows: Based on the set of bidirectional deviation metrics of the equipment, the bidirectional deviation metrics of all circuit breakers and reclosers in the set of bidirectional deviation metrics of the equipment are extracted, sorted in descending order by the comprehensive vulnerability index, and the bidirectional deviation metrics of each equipment and the corresponding equipment number are extracted and recorded in turn to generate a descending sorted sequence of bidirectional deviation metrics of the equipment. Based on the descending sorted sequence of the device bidirectional deviation metric values, the total number of devices in the descending sorted sequence of device bidirectional deviation metric values ​​is counted, and 20% of the total number of devices is used as the screening threshold. Starting from the top of the sorted sequence, the device bidirectional deviation metric values ​​and corresponding device numbers are selected one by one to generate a sequence of high-risk network components.

7. The intelligent alarm system for power distribution automation terminals according to claim 1, characterized in that, The steps for obtaining the dynamic risk heat map of the power distribution terminal are as follows: Based on the network high-risk component sequence, the sorting position of each circuit breaker and recloser in the network high-risk component sequence is read one by one, and according to the correspondence between the sorting position and the preset alarm color level, the corresponding alarm color is marked on the component number position in the power grid wiring diagram or geographic information view to generate a dynamic risk heat map of the power distribution terminal.

Images